Multi-cell SOC estimation for Li-Ion battery applied to an energy storage system

Abstract

Energy Storage Systems (ESS) market is growing exponentially and Lithium-Ion batteries are currently the technology that present most advantages, including cost, power and energy density. A remarkable difference between this kind of battery and the traditional lead-acid is the need for a Battery Management System (BMS), an electronic device responsible for maintaining the battery in a safe range of temperature, current and voltage, among other features. The BMS is equipped with multiple sensors that monitor the ESS and with a processor that runs algorithms to estimate important information for the user, such as the State of Charge (SOC), State of Health (SOH) and End of Life (EOL). This paper focuses on SOC estimation for a multi-cell ESS, ready to be embedded in the BMS and explores a smoothing algorithm to improve the estimation results. The battery equivalent electric circuit, the description of its state space model, the one-step ahead prediction with the Extended Kalman Filter (EKF) and the state smoothing algorithm are presented. To validate the methodology, the results were compared with laboratory tests. The error found by the smoothing algorithm was smaller than the EKF step ahead prediction error.